JP2009507371A - Built-in self-test of heat treatment system - Google Patents

Abstract

  A method of monitoring a heat treatment system (100, 200) in real time using a built-in self test (BIST) table is to place a plurality of wafers (W) in a treatment chamber (202) of the heat treatment system (100, 200). Performing a real-time dynamic model (330) to generate a predicted dynamic process response for the processing chamber (202) in process, creating a dynamic process response by measurement, dynamic by prediction Determining a dynamic estimation error using a difference between the process response and the measured dynamic process response, and determining the dynamic estimation error as an operating threshold established by one or more rules in the BIST table Includes a step to compare with.

Description

  The present invention relates to a fault condition detection, diagnosis and prediction method for a heat treatment system used in a semiconductor process.

  Failure conditions in reactors used in semiconductor manufacturing can result in lost revenue due to waste and non-productive equipment downtime. In this regard, so far, the focus has been on device software that monitors the operation of the device and alerts when unacceptable process deviations occur or other fault conditions are encountered.

  However, what is needed is a way to continuously determine the “health” of a device in order to detect even “newly emerging” fault conditions. Equipment and process engineers are asking the question “how much more can be put into service before service or repair is needed”. Take, for example, a batch furnace system used in semiconductor processes. In this device, some of the items that engineers are aware of are: a) mass flow controller (MFC) drift, b) in-line flow meter (MFM) problems, c) leak rate, and d) heater elements.

  Traditionally, both device manufacturers and chip manufacturers have relied on planned preventive maintenance (PM) of devices. However, this method is based solely on “rules of thumb” derived from average characteristics, such as mean time between failures (MTBF), to address the detection, diagnosis or prediction of individual device fault conditions. I can't.

  The prior art includes several approaches that attempt to detect device problems. One method described in Patent Document 1 relies on analyzing vibration signals of components. However, this approach is limited only to larger mechanical devices with repetitive motion, such as pumps. This approach does not address most of the critical components such as mass flow controllers, heaters, thermocouples, etc.

Another approach described in US Pat. No. 6,057,836 relies on extending a previously known statistical process control (SPC) chart. However, this approach focuses on a single component and does not provide system level detection or diagnosis. Moreover, this approach cannot cope with prediction at all.
US Pat. No. 6,195,621 US Pat. No. 6,351,723

  It is an object of the present invention to provide a method for monitoring a heat treatment system in real time using a built-in self test (BIST) table.

  A method according to one embodiment includes: placing a plurality of wafers at different height positions in a processing chamber of a thermal processing system; starting a process, wherein a process parameter is determined from a first value during the process. Either changed to a second value or maintained at an operating value during the process; real time to generate a predictive dynamic process response for the processing chamber during the process Performing a dynamic model; creating a dynamic process response by measurement with respect to the processing chamber during the process; using the difference between the dynamic process response by prediction and the dynamic process response by measurement, Determining a dynamic estimation error; and the dynamic estimation error with one or more first operating thresholds established by one or more rules in the BIST table; It includes a compare stages. When the dynamic estimation error is not within the operating threshold established by at least one of the rules in the BIST table, the process is paused or stopped.

  In a further embodiment, the dynamic estimation error may be further compared to a warning threshold when not within the operating threshold. When the dynamic estimation error is within the warning threshold, a warning message is sent and the process can continue. Also, when the dynamic estimation error is not within the warning threshold, a fault message is sent and the process can be stopped.

  Many more complete appreciation of the present invention and its attendant advantages will be readily apparent with reference to the following detailed description, particularly when considered in conjunction with the accompanying drawings.

  The semiconductor processing system may include a heat treatment system. According to the present invention, the heat treatment system can be monitored in real time using a BIST table. In one embodiment, one or more process parameters are changed or maintained, a response thereto is predicted and measured, and an estimation error is determined to determine the operating threshold established by the rules in the BIST table and / or Compared to warning threshold. The process can then be continued, paused or stopped depending on whether the estimation error is within the operating threshold and / or warning threshold.

  In one embodiment of the present invention, there is provided a method for detecting, diagnosing and predicting fault conditions to identify faults and error conditions in semiconductor manufacturing equipment and to indicate fluctuations and degradations that can result in fault conditions. Is done.

  In another or further embodiment, the present invention combines techniques from real-time modeling and estimation focused on time series analysis with a rule-based reasoning system to provide a novel for heat treatment systems. Create a “built-in self-test” software application. The heat treatment system can include a chamber that can be a hot wall or a cold wall. These systems can be configured for single wafer processing, multi-wafer processing, or batch processing and are used to process various types of substrates, including silicon wafers and liquid crystal display (LCD) panels. Is possible.

  In another or further embodiment, the present invention identifies faults and / or error conditions in the thermal processing system and also indicates drifts and degradations that can lead to near future fault conditions. Data for analyzing the heat treatment system is obtained either in a “passive mode” in the productive operation of the furnace or in an “active mode” in which a periodic self-test is performed when not in use.

  Next, the present invention will be described with reference to the drawings. FIG. 1 is a schematic perspective view of a heat treatment system according to an embodiment of the present invention. The heat treatment system 100 has a housing 101 that forms an outer wall of the system when the system is placed in a clean room. The inside of the housing 101 is divided into a carrier transport area 107 in which the carrier 102 is carried in and out and held by a partition (bulkhead) 105, and a substrate to be processed (for example, a semiconductor wafer W) in the carrier 102 (for example, a semiconductor wafer W). (Not shown) is divided into a loading area 109 that is transferred to a boat 103 that is mounted on and detached from a vertical heat treatment furnace (chamber) 104.

  As shown in FIG. 1, a carry-in port 106 for introducing and taking out the carrier 102 by an operator or an automatic transfer robot (not shown) is provided on the front surface of the housing 101. The carry-in entrance 106 includes a door that can move in the vertical direction so as to open and close the carry-in entrance 106. A stage 108 on which a carrier is placed is provided in the vicinity of the carry-in entrance in the carrier conveyance area 107.

  As shown in FIG. 1, a sensor mechanism 109 for opening the lid (not shown) of the carrier 102 and detecting the position and number of the semiconductor wafer W is provided on the back surface of the stage 108. . A shelf-like storage unit 110 for storing a plurality of carriers 102 is formed above the stage 108.

  As a table on which the carrier 102 is mounted for transporting the semiconductor wafer W, two carrier arrangement portions (transport stages) 111 are provided at positions separated in the vertical direction. Therefore, the semiconductor wafer W can be transferred to another carrier 102 in the other carrier arrangement part 111 while one carrier 102 is exchanged in one carrier arrangement part 111, so that the throughput of the heat treatment system 100 is increased. It is done.

  In the carrier transport area 107, a carrier transport mechanism 112 for unloading / loading the carrier 102 with respect to the stage 108, the storage unit 110, and the carrier placement unit 111 is disposed. The carrier transport mechanism 112 includes a lift arm 112b that can be moved up and down by a lift mechanism 112a provided on one side of the carrier transport region 107, and a bottom portion of the carrier 102 for transporting the carrier 102 in the horizontal direction. And a transfer arm 112c attached to the lift arm 112b so as to support it.

  For example, the carrier 102 may be a sealed type that can accommodate 13 or 25 wafers and can be sealed by a lid (not shown). The carrier 102 can have a plastic container that accommodates and holds the wafer group W in multiple stages in a horizontal posture with a predetermined pitch apart from each other in the vertical direction. In one embodiment, the diameter of the wafer W may be 300 mm. In other examples, other wafer sizes may be used. The lid (not shown) is detachably attached to the wafer insertion port so that the wafer insertion port formed on the front surface of the carrier 102 can be sealed.

Clean air that has passed through a filter (not shown) is supplied to the carrier transport region 107, and the carrier transport region 107 is filled with clean air. In addition, a clean atmosphere is supplied to the loading area 109, and the loading area 109 is also filled with the clean atmosphere. In another example, an inert gas such as nitrogen (N ₂ ) is supplied to the loading area 109 and the loading area is filled with the inert gas.

  As shown in FIG. 1, the partition 105 has two openings 113 on the upper side and the lower side for transporting the carrier 102. The opening 113 can be aligned with the carrier placement portion 111. Each opening 113 includes a lid (not shown) for opening and closing the opening 113. The opening 113 is formed so that the size of the opening 113 is substantially the same as the size of the wafer insertion opening of the carrier 102, whereby the semiconductor wafer W passes through the opening 113 and the wafer insertion opening from the carrier. Can be put in and out.

  Also, a notch (cutting portion) provided around the semiconductor wafer W for aligning the crystal orientation of the semiconductor wafer W below the carrier placement portion 111 and along the vertical center line of the carrier placement portion 111 is aligned. A notch position alignment mechanism 115 is disposed. The notch position alignment mechanism 115 has an opening on the loading area 109 side. The notch position alignment mechanism 115 is adapted to align the notches of the semiconductor wafer group W transferred from the carrier 102 on the carrier placement unit 111 by the transfer mechanism 122.

  The notch position alignment mechanism 115 has two devices that can align the notches of the wafer group at positions separated in the vertical direction. Thus, the throughput of the thermal processing system 100 is increased because the other apparatus can align the other wafer group while the other apparatus returns the aligned wafer group to the boat 103. Each apparatus is adapted to align a plurality of wafers, eg, 3 or 5, at a time, thereby substantially reducing the time for transporting the wafers.

  The heat treatment furnace 104 is disposed on the back and upper part of the loading area 109. The heat treatment path 104 has a heating furnace opening 104a at the bottom thereof. A lid 117 is provided under the heating furnace 104. The lid portion 117 is adapted to be moved up and down by an elevating mechanism (not shown) in order to mount the boat 103 on the heating furnace 104, to remove it from the heating furnace 104, and to open and close the heating furnace opening 104a. Yes. For example, the boat 103 capable of holding a large number of semiconductor wafers W such as 100 or 150 in a multi-stage shape at equal intervals in the vertical direction is adapted to be disposed on the lid portion 117. The boat 103 is made of quartz or the like. The heat treatment furnace 104 includes a shutter 118 for closing the heating furnace opening 104a when the lid portion 117 is removed after the heat treatment and the boat 103 is taken out at the position of the heating furnace opening 104a. The shutter 118 is adapted to pivot horizontally to open and close the furnace opening 104a. In order to rotate the shutter 118, a shutter drive mechanism 118a is provided.

  Still referring to FIG. 1, a boat arrangement portion (boat stage) 119 for placing the boat 103 when the semiconductor wafer W is taken in and out of the boat 103 is arranged in a side region of the loading region 109. The boat arrangement part 119 has a first arrangement part 119a and a second arrangement part 119b arranged between the first arrangement part 119a and the lid part 117. Adjacent to the boat arrangement portion 119, a ventilation unit for purifying the circulating gas (clean air or inert gas) in the loading region 109 using a filter is arranged. Between the carrier arrangement part 111 and the heat treatment furnace 104 in the lower part of the loading area 109, a boat conveyance mechanism 121 for conveying the boat 103 between the boat arrangement part 119 and the lid part 117 is arranged. . Specifically, the boat transport mechanism 121 includes the first placement portion 119a or the second placement portion 119b and the lowered lid portion 117, and the first placement portion 119a and the second placement portion 119b. The boat 103 is transported between the two.

  Above the boat transport mechanism 121, between the carrier 102 on the carrier placement part 111 and the boat 103 on the boat placement part 119, more specifically, the carrier 102 on the carrier placement part 111 and the notch position alignment mechanism. 115, between the notch position alignment mechanism 115 and the boat 103 on the first arrangement part 119a of the boat arrangement part 119, and on the boat 103 and the carrier arrangement part 111 after the heat treatment on the first arrangement part 119a. A transport mechanism 122 for transporting the semiconductor wafer W is arranged between the carrier and the other carrier.

  As shown in FIG. 1, the boat transport mechanism 121 has an arm 123 that supports one boat 103 in the vertical direction and can move (expand / shrink) in the horizontal direction. For example, the boat 103 can be transported in the radial direction (horizontal linear direction) with respect to the rotation axis of the arm 123 by synchronously rotating the arm 123 and a support arm (not shown). Thus, the area for carrying the boat 103 can be minimized and the width and depth of the heat treatment system 100 can be reduced.

  The boat transport mechanism 121 transports the boat 103 including the unprocessed wafer W from the first placement portion 119a to the second placement portion 119b. The boat transport mechanism 121 transports the boat 103 including the processed wafer W from the lid portion 117 to the first placement portion 119a. The boat transport mechanism 121 transports the boat 103 including unprocessed wafers onto the lid portion 117. Thus, the unprocessed wafer W is prevented from being contaminated by particles or gas coming from the boat 103 including the processed wafer W.

  When the carrier 102 is arranged on the stage 108 through the carry-in port 106, the sensor mechanism 109 detects the arrangement state of the carrier 102. Then, the lid of the carrier 102 is opened, and the sensor mechanism 109 detects the position and number of the semiconductor wafers W in the carrier 102. Then, the lid of the carrier 102 is closed again, and the carrier 102 is transported to the storage unit 110 by the carrier transport mechanism 112.

  The carrier 102 stored in the storage unit 110 is transported onto the carrier placement unit 111 by the carrier transport mechanism 112 in a timely manner. After the lid of the carrier 102 on the carrier arrangement part 111 and the door of the opening 113 of the partition 105 are opened, the semiconductor wafer W is taken out from the carrier 102 by the transfer mechanism 122. Then, the transfer mechanism 122 sequentially transfers the semiconductor wafer groups W to the empty boat 103 arranged on the first arrangement unit 119a of the boat arrangement unit 119 via the notch position alignment mechanism 115. While the wafer W is being transferred, the boat transfer mechanism 121 is lowered so as to retract from the transfer mechanism 122 so that the collision between the boat transfer mechanism 121 and the transfer mechanism 122 is prevented. Thus, the time for transporting the semiconductor wafer W can be shortened, whereby the throughput of the heat treatment system 100 can be substantially increased.

  After the transfer of the wafer W is completed, the transfer mechanism 122 can move laterally from the operating position to the standby position in the area on the other side of the housing 101.

  After the heat treatment is completed, the lid portion 117 is lowered, and the boat 103 and the heat-treated wafer are moved from the heating furnace 104 to the loading region 109. Immediately after the lid 117 is separated from the boat 103, the shutter 118 seals the opening 104a of the heating furnace. This minimizes heat transfer from the heating furnace 104 to the loading area 109 and minimizes heat transferred to equipment in the loading area 109.

  When the boat 103 containing the processed wafers W is transported out of the heating furnace 104, the boat transport mechanism 121 moves another boat 103 including the unprocessed wafers W from the first placement section 119a to the second. Transport to placement section 119b. The boat transport mechanism 121 transports the boat 103 containing the processed wafers W from the lid portion 117 to the first placement portion 119a. Then, the boat transport mechanism 121 transports the boat 103 including unprocessed wafers from the second arrangement portion 119b onto the lid portion 117. Therefore, when the boat 103 is moved, the unprocessed semiconductor wafer W in the boat 103 is prevented from being contaminated by particles or gas coming from the boat 103 including the processed wafer W.

  After the boat 103 including the unprocessed wafer W is transported onto the lid portion 117, the boat 103 and the lid portion 117 are introduced into the heating furnace 104 through the opening 104a after the shutter 118 is opened. Then, the boat 103 including the unprocessed wafer W is heat-treated. In addition, after the boat W including the processed wafer W is transported onto the first arrangement unit 119a, the processed semiconductor wafer W in the boat 103 is transferred from the boat 103 to the empty carrier 102 on the carrier arrangement unit by the transfer mechanism 122. Returned to. Then, the above cycle is repeated.

  Setup, configuration and / or operational information may be recorded by the thermal processing system 100 or obtained from an operator or another system, such as a factory system. BIST rules and BIST tables can be used to specify actions taken for normal processing and actions taken under exceptional conditions. A setting screen can be used to define and manage BIST rules. BIST rules are stored and can be updated as needed. Documentation and help screens can be provided on how to create, define, assign, and manage BIST rules.

  BIST rules can be used to determine when a process is paused and / or stopped and what happens when the process is paused and / or stopped. BIST rules can also be used to determine when and how to change a process. Furthermore, the rules can be used to determine when to select different dynamic / static models and how to create BIST rules in the process. In general, BIST rules allow system behavior to change based on the dynamic state of the system.

  In one embodiment, the thermal processing system 100 can have a system controller 190 that can include a processor 192 and a memory 194. Memory 194 is coupled to processor 192 and may be used for storing information and instructions to be executed by processor 192. In other examples, different controller configurations may be used. The system controller 190 can also have a port 195 that can be used to couple the thermal processing system 100 to another system (not shown). In addition, the controller 190 may have input and / or output devices (not shown) for coupling the controller 190 to other elements of the system.

  In addition, other elements of the system may include a processor and / or memory (not shown) for executing and / or storing information and instructions to be executed during processing. For example, the memory can be used to store temporary variables or other intermediate information during execution of instructions by various processors in the system. It is also possible for one or more of the system elements to have means for reading data and / or instructions from a computer readable medium. Further, one or more of the system elements may have means for writing data and / or instructions to a computer readable medium.

  The storage device retains computer-executable instructions programmed in accordance with the teachings of the present invention and stores at least one computer-readable data structure, table, record, rule, or other data described herein. Media or memory may be included. System controller 190 may use data from computer readable media memory to generate and / or execute computer-executable instructions. The thermal processing system 100 performs some or all of the methods according to the present invention in response to the system controller 190 executing one or more instruction sequences comprising one or more computer-executable instructions stored in memory. It can be carried out. Such instructions may be received by the controller from another computer, a computer readable medium, or a network connection.

  The present invention is stored on any computer readable medium or combination thereof to control the thermal processing system 100, drive one or more devices to implement the present invention, and the thermal processing system 100 is a human user and And / or includes software that allows interaction with another system, eg, a factory system. This software may include, but is not limited to, device drivers, operating systems, development tools, and application software. The computer-readable medium further includes a computer program product according to the present invention that performs all or part of the processing performed in practicing the present invention (if processing is distributed).

  Further, at least one of the components of the thermal processing system 100 can include a graphic user interface (GUI) element (not shown) and / or a database element (not shown). In alternative embodiments, GUI elements and / or database elements are not required. The system user interface may be implemented on the web and may provide an indication of system status and alert status. For example, a GUI element (not shown) allows the user to: visually check the status; create and edit SPC charts; alerts to more efficiently solve, diagnose, and report problems with the thermal processing system 100 View data; configure data collection application; configure data analysis application; examine historical data and review latest data; generate alert emails; launch multivariate model; view diagnostic screen And provide an easy-to-use interface that allows viewing / creating / editing BIST rules and tables.

  FIG. 2 illustrates a portion of a heat treatment system 200 according to an embodiment of the present invention, partially cut away. In the illustrated embodiment, a heating furnace system 205, an exhaust system 210, a gas supply system 260, and a controller 290 are shown.

  The heating furnace system 205 includes a vertical processing chamber (reaction tube) 202 having a double structure including an inner tube 202a and an outer tube 202b made of quartz, for example, and a metal disposed at the bottom of the processing chamber 202 And a cylindrical connecting pipe 221. The inner tube 202a has an interrupted and open top and is supported by a connecting tube 221. The outer tube 202b has a top that is closed without interruption, and a lower end that is airtightly sealed to the upper end of the connecting tube 221.

  In the processing chamber 202, a large number of wafers W (for example, 150 wafers) are horizontally loaded on a wafer boat (wafer holder) 223 in a shelf shape at a constant pitch up and down. The wafer boat 223 is held on a lid portion 224 via a heat insulating cylinder (heat insulating material) 225, and the lid portion 224 is coupled to the movable means 226. The furnace system 205 can also include a heater 203 in the form of, for example, a resistor disposed around the processing chamber 202. The heater 203 may be composed of five stages of heaters 231-235. In other examples, different heater configurations may be used. Each heater stage 231-235 is powered independently from each other by a power controller 236-240 coupled thereto. The heater stages 231-235 can be used to divide the interior of the processing chamber 202 into five compartments.

  A gas supply system 260 coupled to the controller 290 and the furnace system 205 is shown. The connecting pipe 221 has a plurality of gas supply pipes 241 to 243 for sending gas into the inner tube 202a. In one embodiment, dichlorosilane, ammonia, and nitrogen are fed into respective gas supply tubes 241, 242, 243 via flow regulators 244, 245, 246, such as, for example, a mass flow controller (MFC). In other examples, other process gases may be used.

  An exhaust pipe 227 is connected to the connecting pipe 221 for exhausting through a gap between the inner pipe 202a and the outer pipe 202b. The exhaust pipe 227 is connected to an exhaust system 210 that may include a vacuum pump (not shown). In order to adjust the pressure in the processing chamber 202, a pressure regulator 228 including a valve, a butterfly valve, a valve driver and the like is inserted into the exhaust pipe 227.

  The furnace system 205 can also have a number of sensors. In the illustrated embodiment, five internal temperature sensors (thermocouples) 251-255 are arranged in the vertical direction inside the inner tube 202a. The internal temperature sensors 251 to 255 are covered with, for example, a quartz tube in order to prevent metal contamination of the semiconductor wafer W. The internal temperature sensors 251 to 255 are arranged so as to correspond to the above five sections. In other examples, a different number of compartments and a different number of internal temperature sensors may be used, and the internal sensors may be arranged differently. In another embodiment, optical techniques can be used to measure temperature.

  On the outside of the outer tube 202b, a plurality of external temperature sensors (thermocouples / thermometers) 261-265 are arranged in the vertical direction. External temperature sensors 261-265 can also be arranged to correspond to the above five sections. In other examples, a different number of compartments and a different number of external temperature sensors may be used, and the external sensors may be arranged differently.

  The controller 290 can be used to control process parameters such as, for example, the temperature of the process atmosphere, the gas flow rate, and the pressure within the process chamber 202. The controller 290 receives output signals from the internal temperature sensors 251 to 255 and the external temperature sensors 261 to 265, and outputs control signals to the power controllers 236 to 240, the pressure regulator 228, and the flow regulators 244 to 246.

  Configuration, configuration and / or operational information can be recorded by the controller 290 or obtained from an operator or another controller, such as the controller 190 (FIG. 1). The controller 290 can also use BIST rules and BIST tables to determine actions taken for normal processing and actions taken under exceptional conditions. A setting screen can be used to define and manage BIST rules. The controller 290 can store BIST rules and update them as needed. Documentation and help screens can be provided on how to create, define, assign, and manage BIST rules.

  The controller 290 can use BIST rules to determine when a process is paused and / or stopped and what happens when the process is paused and / or stopped. BIST rules can also be used to determine when and how to change a process. Furthermore, the rules can be used to determine when to select different dynamic / static models and how to create BIST rules in the process. In general, BIST rules allow system behavior to change based on the dynamic state of the system.

  In one embodiment, the controller 290 may include a processor 292 and a memory 294. Memory 294 is coupled to processor 292 and may be used for storing information and instructions to be executed by processor 292. In other examples, different controller configurations may be used. The system controller 290 can also have a port 295 that can be used to couple the controller 290 to another computer and / or network (not shown). In addition, the controller 290 can have input and / or output devices (not shown) for coupling the controller 290 to the furnace system 205, the exhaust system 210, and the gas supply system 260.

  Controller 290 can include means for reading data and / or instructions from a computer-readable medium. Further, the controller 290 may include means for writing data and / or instructions to a computer readable medium.

  Memory 294 retains computer-executable instructions programmed in accordance with the teachings of the present invention and stores data structures, tables, records, rules, or other data described herein, at least one computer-readable medium or memory. Can be included. Controller 290 may use data from computer-readable media memory to generate and / or execute computer-executable instructions. The furnace system 205, exhaust system 210, gas supply system 260, and controller 290 are responsive to executing one or more instruction sequences comprising one or more computer-executable instructions stored in memory. A part or all of the method can be performed. Such instructions may be received by the controller from another computer, a computer readable medium, or a network connection.

  The present invention may be stored on any computer readable medium or combination thereof to control the furnace system 205, the exhaust system 210, the gas supply system 260 and the controller 290 to implement one or more devices to implement the present invention. And software that allows one or more of the system elements to interact with a human user and / or another system. This software may include, but is not limited to, device drivers, operating systems, development tools, and application software. The computer-readable medium further includes a computer program product according to the present invention that performs all or part of the processing performed in practicing the present invention (if processing is distributed).

  The controller 290 can have GUI elements (not shown) and / or database elements (not shown). In alternative embodiments, GUI elements and / or database elements are not required. The system user interface may be implemented on the web and may provide an indication of system status and alert status. For example, a GUI element (not shown) allows the user to: view the status; create and edit the chart; view the alert data to solve, diagnose, and report problems more efficiently; Configure data collection application; configure data analysis application; examine historical data and review latest data; generate alert emails; view / create / edit / run dynamic and / or static models Browsing diagnostic screens; and providing an easy-to-use interface that allows BIST rules and tables to be viewed / created / edited.

  During the first processing time, the controller 290 can change the pressure in the processing chamber 202 from one pressure to another. A real-time dynamic pressure model for this particular system configuration can be constructed based on the type of vertical wafer boat 223, the type, location and quantity of wafers W, the type of thermal processing chamber 202, and the recipe to be performed.

  A real-time dynamic pressure model can be implemented to produce a predicted dynamic pressure response for the processing chamber at the first processing time. Also, a measured dynamic pressure response is created for the processing chamber at the first processing time, and using the difference between the predicted dynamic pressure response and the measured dynamic pressure response, A reasonable estimation error can be determined. Furthermore, the dynamic estimation error can be compared to an operational threshold established by one or more rules in the BIST table. When the dynamic estimation error is not within the operating threshold established by at least one of the rules in the BIST table, the process can be stopped and the dynamic estimation error is at least one of the rules in the BIST table. When within the operating threshold established by, the process can continue.

  During the second processing time, the controller 290 can change the temperature in the processing chamber from one temperature to another. For example, the controller 290 may execute a heater control model designed to estimate the temperature of the wafers W loaded in the wafer boat 223 in each compartment. The controller 290 can compare the estimated temperature with the measured temperature, and can correct the estimated temperature of the wafer W based on the comparison result. A technique for controlling a heating device using a model is disclosed in US Pat. No. 6,803,548 filed on Sep. 13, 2001, entitled “Batch-type Heat Treatment Apparatus and Control Method for the Batch-type Heat Treatment Apparatus”. It is disclosed in the specification. This document is incorporated herein by reference.

  One or more real-time dynamic temperature models can be implemented to produce a predicted dynamic temperature response for the processing chamber 202 at the second processing time. Also, for the processing chamber 202 at the second processing time, one or more measured dynamic pressure responses are created, using the difference between the predicted dynamic temperature response and the measured dynamic temperature response. A dynamic estimation error can be determined. Furthermore, the dynamic estimation error can be compared to an operational threshold established by one or more rules in the BIST table. When the dynamic estimation error is not within the operating threshold established by at least one of the rules in the BIST table, the process can be stopped and the dynamic estimation error is at least one of the rules in the BIST table. When within the operating threshold established by, the process can continue.

  Controller 290 also measures using data from internal temperature sensors 251-255, external temperature sensors 261-265, power controllers 236-240, pressure regulator 228 or flow regulators 244-246, or combinations thereof. Static and / or dynamic responses can be created.

  During operation, the temperature of the wafer W in each compartment is estimated and an adaptive technique is used to control the heater stages 231-235 so that the corrected wafer temperature is equal to the temperature indicated by the recipe. . When the temperature rise is complete, adaptive control is used to maintain the temperature of each compartment.

  When the temperature in the processing chamber 202 is stabilized during the third processing time, a process gas is sent into the processing chamber 202, a film formation process is started, and the temperature of the wafer W in each compartment is changed to a temperature recipe. The temperature control is managed so as to be substantially equal to the set temperature. Wafers W in different compartments can be processed at different temperatures. However, the model and recipe have values adjusted so that a relatively uniform thickness film is deposited on different sides of the wafer W.

  One or more real-time dynamic temperature models can be implemented to produce a predicted dynamic process response for the processing chamber 202 at the third processing time. The real-time dynamic model generates the predicted dynamic gas flow response, the predicted dynamic temperature response, or the predicted dynamic pressure response, or a combination thereof, for the processing chamber 202 at the third processing time. Can be realized. Also, for the process chamber 202 at the third process time, one or more measured dynamic pressure responses are created, using the difference between the predicted dynamic temperature response and the measured dynamic temperature response. A dynamic estimation error can be determined. Furthermore, the dynamic estimation error can be compared to an operational threshold established by one or more rules in the BIST table. When the dynamic estimation error is not within the operating threshold established by at least one of the rules in the BIST table, the process can be stopped and the dynamic estimation error is at least one of the rules in the BIST table. When within the operating threshold established by, the process can continue.

  When the film growth is completed, the supply of the film forming gas is stopped and the inside of the processing chamber 202 is cooled. During cooling, the temperature of the wafer W is estimated as necessary, and the estimated temperature is corrected. When the processing is completed, the processed wafer boat 223 is extracted.

  During processing, products such as wafers or LCD substrates are processed by a semiconductor processing system and can be moved from one processing element to another.

  FIG. 3 shows a simplified block diagram of a heat treatment system according to an embodiment of the present invention. In the illustrated embodiment, the processing system 300 includes a system 310, a controller 320, a dynamic model 330, and a comparator 340. In addition, an actuation variable (AV) is shown. These are variables that have a fixed setpoint (SP) in the recipe, or variables that are generated in real time by the controller based on the setpoint in the recipe, such as heater power, mass flow, and exhaust valves Is the angle.

  Two types of process variables (PV) are illustrated which are process states within the device as a result of operating variables. Examples of process variables include paddle or water temperature, reactant concentration at the substrate location, and film thickness on the substrate. Process variables can be categorized as measured process variables (MPV) that are measured using sensors and general process variables (GPV) that are not measured by sensors. Some of the measured process variables can be controlled directly by the controller, these are called controlled process variables (CPV).

  AV, MPV and SP are obtained in real time. By definition, GPVs are not obtained (not measured) and their effects can only be estimated by end-of-run measurements. For example, in a batch furnace, real-time data is obtained for chamber pressure, mass flow and their set points, valve angle, and chamber temperature for each compartment.

  An error condition may occur during system operation. For example, the type of error condition is close to the active component, passive component, or software component failure when the requested task cannot be performed; the active component, passive component, or software component is degraded and the degraded performance is not corrected Component degradation may occur in the event of a failure in the future; and configuration errors may occur when active components, passive components or software elements are not properly configured.

  A thermal processing system and / or system component is in operation, a processing state in which the system and / or system component is processing a wafer and / or substrate, and one or more of the system and / or system component is processing a wafer and / or substrate. It can be in one of two states: a dormant state waiting to do. In an alternative embodiment, a maintenance state in which one or more of the system and / or system components are offline for maintenance, calibration, and / or repair time may also be used.

  These two states provide a unique opportunity to perform BIST related operations. In passive mode, during the processing state, one or more of the processes can “observe” the real-time system response, but it cannot change the processing state. For example, system response can be observed and processing and error conditions can be determined. Furthermore, a dynamic model and / or a static model can be executed, and BIST rules can be changed or new BIST rules can be created.

  In the rest state, no product substrate is being processed, so one or more of the processes can change the processing state. During the dormant mode, one or more of the processes can create the necessary processing state to detect and diagnose degradation and failure conditions. These degraded and fault conditions can be used to create and / or modify one or more BIST rules. Specifically, one or more of the processes may select process parameters so as to exaggerate error conditions rather than hiding or compensating for error conditions. In the dormant mode, the system can be operated using production conditions, or the system can be operated using non-production conditions (eg, no wafer in chamber, no reactive gas flow, etc.). Also, one level of testing can be performed at the part level, and these tests can be static or dynamic. For example, the resistance of the heater can be monitored for any signs of degradation.

  Static measurement is a way to detect errors. However, drift and degradation can have an overall effect on system performance while being small for multiple parameters.

  The dynamic performance of the system is a composite reflection of a set of system parameters that can depend on many variables. For example, the thermal response of the system can be a function of active components, passive components or software elements. Deviations in one or more of these parts can cause errors. For better detection and diagnosis, a “system level” approach can be used.

  Recipes can be used at the system level, a typical recipe gives a setpoint (SP) of measurement process variables (MPV) including controlled variables (CPV), and the system controller gives error = SP-CPV As described above, the operating variable (AV) can be controlled to reduce the error between CPV and SP.

Rules can also be used to specify CPV tolerances in certain critical processes. A warning is issued if the CPV deviates from this range. For example:
CPV-Lower limit (LB)> Warning action threshold (1)
Upper limit (UB) -CPV> Warning action threshold (2)
It is.

  However, this approach only determines the “capability” of the controller to keep the error small, not whether the device is functioning properly. In particular, this approach does not deal with what happens because of general process variables and hence end-of-execution parameters.

In one embodiment, BIST rules are used to determine whether a system and / or system component is “as designed” from real-time data and a dynamic model of the system component. The dynamic model provides a “as designed” system response and can be used to detect error conditions. For example, the error can be calculated using the difference between the modeled response (IPV) and the measured response,
Error = IPV-MPV
Can be calculated as follows.

  If this error is greater than a predetermined operating threshold, a warning is issued. The operational threshold may be part of the BIST table.

  In “active mode”, BIST can be used to create a state that exaggerates the error between the modeled response and the measured response in the proper range of PV. In this dynamic state, the deviation between the model and the measurement is amplified. Under static or stationary conditions, the error appears as a “bias” error and can be hidden by measurement noise. Active mode can significantly increase the probability of an accurate warning.

  For example, in the case of a batch furnace, the model used for BIST is related to various real-time data sets collected from the furnace and includes setpoint, MFC flow, 0-10 Torr and 0-1000 Torr standards. Includes measurement, valve angle, and pressure control parameters.

  These models include both dynamic (transient) behavior and steady state behavior. For example, dynamic behavior models the pressure increase rate. To further describe the mathematical model, the notation shown in Table 1 can be used.

チャンバー圧の昇圧（変化）速度である“ｐドット”（ｐの時間微分係数）は、ガス流及びバルブ角度の関数として、

とモデル化され得る。

The “p dot” (time derivative of p), which is the rate of increase (change) in chamber pressure, is a function of gas flow and valve angle.

And can be modeled.

とすることができる。このモデルが与えられると、ＭＦＣ流量を個々に推定する一手法は、

となる。また、定常的な挙動は、

とモデル化されることができる。このモデルは、残りのパラメータを所与としてパラメータを推定するために使用され得る。すなわち、

である。 A special condition exists when the valve is completely closed, in which case the pressure increase rate does not depend on the pressure controller. In this state, the chamber pressure increase rate p dot is

It can be. Given this model, one way to estimate MFC flow individually is:

It becomes. The steady behavior is

And can be modeled. This model can be used to estimate the parameters given the remaining parameters. That is,

It is.

  Hence, these models provide estimates of multiple process parameters. Inference logic finds a set of consistent explanations for the estimated parameters to make a diagnosis.

  In one embodiment, a method for capturing the dynamic and steady state behavior of a furnace gas flow system can be constructed. The method includes performing a “model development” recipe in the furnace. This recipe is designed to operate under a variety of conditions: 1) Dynamic response of the system under automatic pressure control (APC); 2) Under manual valve control (MVC) 3) Base pressure when the valve is fully open; and 4) Leak rate when the valve is fully closed.

In a test set to determine system response under APC, the pressure set point can be changed and the change in chamber pressure at various gas flow rates can be analyzed. For example,
1) Step change of pressure set point from 0 to 3 Torr,
2) a step change from 3 to 6 Torr, and 3) a step change from 6 to 9 Torr,
Such changes are made.

  When the pressure set point is changed from 0 to 3 Torr, the pressure controller adjusts the opening of the main valve to achieve the target pressure. Initially, the action of this controller completely closes the valve. Under these conditions, the pressure increase rate is a function of the gas flow rate without depending on the valve angle. Typical results are shown in FIGS. FIG. 4 shows the transient behavior of the chamber pressure for changing the pressure set point from 0 to 3 Torr. FIG. 5 shows the transient behavior of the chamber pressure for changing the pressure set point from 3 to 6 Torr. FIG. 6 shows the transient behavior of the chamber pressure for changing the pressure set point from 6 to 9 Torr.

  These figures include three stepped pressure responses, which are the pressure increase rates measured at gas flow rates of 200, 250 and 300 sccm, respectively. This data can be used to obtain a straightforward mathematical model of the pressurization rate versus gas flow rate. For example, a linear relationship y = ax + b can be used where y = pressure increase rate (mTorr / s) and x = gas flow rate (sccm). The results of Table 2 were obtained by the first least squares fit.

これらの結果は図７にも示されており、そこでは、上記の３つの階段状応答の試験条件でのガス流量の関数として昇圧速度が示されている。

These results are also shown in FIG. 7, where the pressure increase rate is shown as a function of gas flow rate under the three step response test conditions described above.

  The parameter “b” should ideally be equal to zero. That is, the pressure increase rate should be zero when there is no gas flow. The measured value is a very small value of about 1 sccm. The main parameter is “a”. As is apparent from the data in Table 2, the boost rate can be modeled in this operating range with a single parameter set consisting of parameters “a” and “b” independent of gas flow rate, which is One of the partial models used in BIST software. For example, when the estimated gas flow rate is approximately equal to 258.6 sccm, the measured pressurization rate is approximately equal to 75 mTorr / s.

を用いて得られる。 The system response under MVC can be determined and the relationship of steady state pressure p as a function of gas flow rate and valve angle is

Is obtained.

  This function is expected to be very non-linear (unlike the linear function obtained with respect to the boost rate). In one example, specific values of this function are obtained experimentally under a combination of the following conditions: MFC flow rates of 200, 250 and 300 sccm, and valves of 3.3, 4.3 and 5.3% angle. For example, values obtained with one system are shown in Table 3, and chamber pressure as a function of flow rate at various valve angles is illustrated in FIG.

一実施形態において、実時間推定手法が使用され得る。他の例では、その他の手法が用いられてもよい。半導体処理システム、及び／又はシステム部品又はサブシステム部品の１つ以上は、次の非線形微分方程式“ｘドット”と出力方程式ｙとの組を用いてモデル化されることができる。

In one embodiment, a real time estimation approach may be used. In other examples, other techniques may be used. The semiconductor processing system and / or one or more of the system components or subsystem components can be modeled using the following set of nonlinear differential equations “x dots” and output equations y.

ただし、ｘは温度、圧力及び反応物状態からなり得る状態ベクトルであり；ベクトルｐは、例えば熱容量、熱伝導率及び速度定数などのモデルパラメータから成り；ベクトルｕは、例えばヒータ電力などのプロセスに与えられる入力から成り；ｗは、平均値ゼロを有し、Ｅ（ｗ）＝０（Ｅ（）は期待値演算子を表す）の付加白色雑音であり；νは、平均値ゼロを有し、Ｅ（ν）＝０の付加白色雑音である。

Where x is a state vector that can consist of temperature, pressure and reactant state; vector p consists of model parameters such as heat capacity, thermal conductivity and rate constant; Consists of given inputs; w has an average value of zero and E (w) = 0 (E () represents the expectation operator) is an additive white noise; ν has an average value of zero , E (ν) = 0 additional white noise.

Given an initial state x ₀ , an input u, and a parameter p, the above differential equation is complete to calculate the evolution of the state.

ただし、ｋは時間インデックスである。初期状態の共分散Ｐ_０は、Ｔを転置（transposition）演算子として、Ｅ｛ｘ_０ｘ_０ ^T｝＝Ｐ_０である。 The model can also be linearized for real-time applications. This linearization may result in one or a set of models that describe system behavior along a nominal trajectory. These linear models are represented in the form of state space by matrices A _i , B _i and C _i for the i th time interval. Hence, the nonlinear model is replaced by a series of discrete-time linear models:

Here, k is a time index. Covariance _P initial state _0, the transposition T (transposition) operator, is ^{_{E {x 0 x 0 T}}} = P 0.

を与える。ただし、“＾”は推定値を指し示している。 A convenient approach to construct a real-time estimation means is to use a Kalman (Kalman) filter matrix L _{i. Li} is

give. However, “^” indicates an estimated value.

となる。 When performing a steady state check, a dynamic model of the system (without the noise term) can be used, and the set of nonlinear differential equation xdot and output equation y is

It becomes.

In the steady state, it x dot = 0, state, steady-state values of the input and output are respectively _x s, _{u s,} and _{y s.} Using known steady state values (eg, u _ref and y _ref ) of the input and output of the reference system, the steady state value of a given system can be monitored and compared to this reference value. Specifically, if the feedback controller drives the system output y _ref, the value of u _s may be inspected. That is,
Drive the output to the reference value y _s → y _ref , and therefore ‖y _s −y _ref ‖ ≦ ε,
Check the input against the reference value ‖u _s −u _ref ‖ ≦ ε?
However, ε is selected to be a sufficiently small value for a specific system.

  If this difference is small, it is indicated that the system under test (SUT) is operating like a reference system, otherwise it may be in an error state.

  For example, consider a reference furnace system having five compartments. Assume that it is determined that the heater power should be as shown in Table 4 when all compartments are 600 ° C. on the reference furnace.

同一の５個の６００℃の温度区画にあるＳＵＴ加熱炉が、第５区画のヒータ電力が２６００Ｗであることを報告するとき、何らかの種類のエラー状態が指し示されていることは明らかである。このエラー状態はヒータ区画やＯリング等々に伴うものであることが考えられ、更なる試験で診断されなければならない。

It is clear that some kind of error condition is indicated when the SUT furnaces in the same five 600 ° C. temperature zones report that the heater power in the fifth zone is 2600 W. This error condition can be associated with heater compartments, O-rings, etc. and must be diagnosed in further tests.

で与えられるような基準システムで作り出された線形推定手段を考える。 In one embodiment, a dynamic response test can be performed. As described herein, one approach to detecting errors is to monitor the dynamic response of the system and compare it to a reference. For example, dynamic real-time estimation means are used for this purpose. The output estimate is

Consider a linear estimator created with a reference system such as

が合理的に小さいかどうか、基準システムの推定値を用いてＳＵＴ出力を検査する。 When the SUT is producing output y _k , whether these two are close enough, ie the following values:

Check the SUT output using an estimate of the reference system to see if is reasonably small.

  Hence, these models provide estimates of multiple process parameters. Inference logic finds a set of consistent explanations for the estimated parameters to make a diagnosis.

  The present invention provides a method for capturing the dynamic and steady state behavior of a furnace gas flow system. The method includes performing a “model development” recipe in the furnace. This recipe is designed to operate under a variety of conditions to generate a large number of BIST rules.

  The main requirements for processing wafers are tight critical dimension (CD) control, tight profile control, and tight uniformity control, both within and between wafers. For example, variations in CD measurement, profile measurement, and uniformity measurement are caused by variations in temperature profiles across wafer compartments and variations in thermal response between wafers.

  In general, raw silicon wafers are relatively flat and are manufactured within strict specifications. However, multiple films are deposited on the wafer during multiple heat treatments so that the wafer can be significantly curved. Wafer curvature can adversely affect CD uniformity by causing problems during processing such as lithography and development processes.

  A BIST system can be used to detect a wafer position error and retract it when the wafer has excessive curvature. The BIST system can use real-time data from one or more measurement devices in a “mathematical model” to estimate and / or compensate for wafer curvature. This model may be static or dynamic, and may be linear or non-linear.

  In one embodiment, the BIST system is for detecting, diagnosing and / or predicting system performance when flat and / or warped wafers are placed in and / or processed in a thermal processing chamber. Use the dynamic response of the chamber. For example, wafers with different curvatures produce different dynamic thermal responses when they are placed in and / or processed within a thermal processing chamber.

  In one embodiment, an “active” approach is used. In this case, a combination of processing parameters (eg, control mode, temperature control section set point, chamber section power) can be actively changed, and the resulting dynamic response of the temperature field is determined by the curvature of the wafer. Can be used to perform estimation and monitoring. The amount of change in the processing parameters is intended to provide information regarding the curvature of the wafer.

  Various active techniques can be used. In the first active example, one or more test wafers are loaded and / or processed to create different dynamic states of the thermal system. Thereby, a dynamic model of the thermal response can be created and compared with a pre-calculated thermal model to estimate system performance. In other examples, multiple dynamic models may be activated in real time, and the estimated error between them and the measured thermal response is examined to estimate the curvature of the wafer. For example, the plurality of dynamic models may be created in advance using real time data from a reference system. In a third active example, thermal response variations, including peak variations in the temperature field between compartments, may be examined. In the fourth example, in addition to the above-described peak variation value, the timing of these peak variations may be checked.

  Also, various measurement methods can be used in the “active” mode. For example, the dynamic response of the temperature field can be measured to estimate and monitor system performance. For example, the thermal response of each compartment may be measured and thermal response variations including peak temperature variations between the compartments may be examined. In another example, in addition to the above-described peak variation values, the timing of these peak variations may be checked. In a third example, multiple dynamic models may be activated in real time, and the estimation error between them and historical data may be examined to estimate system performance. For example, the history data may be created in advance using real time data from the system.

  Other embodiments may also be designed for both real-time and non-real-time comparisons. In a real-time approach, system performance can be estimated and monitored in real time during wafer processing. In a non-real time approach, data is processed at a later time and system performance is estimated and monitored after one or more of the wafers are processed. In other embodiments, virtual sensors may be used to “measure” wafer temperature in real time and eliminate the need to use wafers with measurement means during production. For example, a virtual sensor is a component of a dynamic model or a real-time model, a physical sensor component that measures physical variables such as temperature, a tuned variable that controls variables such as applied voltage or power to a heater There may be software algorithm elements that associate the components and components of the dynamic model with information from physical sensors and tuned variables. This virtual sensor may be viewed as a composite device that integrates information from multiple “physical” sensors based on an algorithm. This virtual sensor is an adaptive device that can provide historical data, real-time data, and predictive data.

FIG. 9 schematically illustrates one embodiment of a dynamic model that characterizes one or more of the responses of a heat treatment system according to an embodiment of the present invention. In the illustrated embodiment, four nodes or model elements (M ₁ , M ₂ , M ₃ and M ₄ ) 948, 950, 952, 954 are shown. However, in other embodiments of the present invention, a different number of model elements may be used and the model elements may be arranged in different architectures.

The dynamic model 904 also receives control inputs (U) 962, such as heater power, chamber pressure, gas flow, and wafer information. The model also receives a disturbance input (D) 956, such as an unmeasured variation. This model determines a controlled output (Z) 958 such as wafer temperature and a measured output (Y) 960 such as chamber temperature. The model structure can be expressed as Z = M ₁ U + M ₃ D and Y = M ₂ U + M ₄ D. In other examples, different representations of the model structure may be used.

Dynamic model 904 tracks the “state” of the system and associates input 962 with outputs 958, 960 in real time. For example, U, Y can be measured, and by using the dynamic model 904, D is estimated using Y = M ₂ U + M ₄ D _est and Z is Z _est = M ₁ U + M ₃ D _est Can be estimated. However, the subscript “est” represents estimation.

  When creating a dynamic model 904, wafer position, wafer curvature, and chamber effects can be incorporated into the model. For example, the dynamic model 904 uses a first principle model based on heat transfer, gas flow and reaction kinetics, or an online model created using real-time data collected from a processing system such as a heat treatment system. Can be created.

  During model development, the first principle model may be numerically implemented on a suitable microprocessor within a suitable software simulation application such as Matlab. This software application resides in a suitable electronic computer or microprocessor that is operated to make an approximation of physical performance. However, other numerical approaches are also contemplated by the present invention.

  To model the amount of heat, a model-based linear or non-linear multivariable control technique in which the controller comprises a mathematical model of the controlled system may be used. This multivariable controller may be based on any of the recent control design methods such as the linear quadratic Gaussian (LQG) method, the linear quadratic regulator (LQR) method, and the H-infinity method. This heat quantity model may be either linear or non-linear, and may be either SISO or MIMO. A multivariable control approach (ie, MIMO) considers all inputs and their impact on outputs. Several other techniques for modeling the amount of heat are also available, such as physical models and data driven models.

  FIG. 10 is a simplified flow diagram illustrating a method for monitoring a thermal processing system in real time using a built-in self test (BIST) table according to an embodiment of the present invention.

  A BIST table can be constructed for heat treatment systems, temperature control components, pressure control components, gas supply components, machine components, computational components or software elements, or combinations thereof.

  At step 1010, one or more pre-processing processes are performed on a group of wafers disposed in a processing chamber of the thermal processing system. The wafer groups are arranged at different heights in the processing chamber, and the processing chamber is sealed. The wafer group may include a product wafer, a wafer with measurement means, a test wafer or a dummy wafer, or a combination thereof. In other examples, there need not be multiple wafers. For example, a vertical boat is used to place a group of wafers in a heating furnace.

  During the pre-processing process, an operating state is established. For example, chamber pressure, chamber temperature, substrate temperature, and / or process gas conditions can be changed to operating values during pretreatment.

  The heat treatment system obtains pretreatment data and uses the pretreatment data to construct one or more recipes for use in the pretreatment process. The pretreatment data may also include dynamic modeling information and / or static modeling information for predicting the performance of the heat treatment system during the pretreatment process. These data may further include data measured and / or predicted from previous runs.

  The preprocessing data may include recipe data, history data, and wafer data. The wafer data is, for example, critical dimension (CD) data, profile data, thickness data, uniformity data, and optical data such as refractive index (n) data and attenuation coefficient (k) data. The wafer data may also include the number of layers, layer position, layer composition, layer uniformity, layer density, and layer thickness. The layers can include semiconductor materials, resist materials, dielectrics, and / or metals. Further, the data may include correction data, error data, measurement data or history data, or a combination of two or more thereof.

  During the pre-processing process, a process may be performed in which one or more of the processing parameters are changed from a first value to a second value. There are one or more real-time dynamic models for preprocessing that can be used to predict the system response when the processing parameter is changed from the first value to the second value. This model may be based on the type of wafer boat, the type of wafer, the location and quantity, the type of thermal processing chamber, and / or the recipe to be performed. For example, the plurality of wafers described above can include curved wafers and / or wafers with measurement means, and the model can take into account factors such as wafer curvature. When a curved wafer is placed in the boat, the gap between the two curved wafers is made variable, and the heat transfer and gas flow are also made variable.

  During the pre-processing process, a real-time dynamic model for pre-processing is executed to generate a predicted dynamic pre-processing process response. For example, a real-time dynamic model for preprocessing can predict dynamic preprocessing process response when one or more of the processing parameters are changed from a first value to a second value during the preprocessing process. Can be executed. This dynamic model can be executed before, during or after the pre-processing process is executed. One or more different dynamic models for preprocessing may be executed when the preprocessing process is executed multiple times and / or when multiple processes are executed.

  A dynamic pretreatment process response with one or more measurements is created during the pretreatment process. A dynamic pre-processing process response by measurement can be created each time one or more of the processing parameters are changed from a first value to a second value during the pre-processing process. When the pretreatment process is performed multiple times, a dynamic pretreatment process response with one or more different measurements can be created.

  In addition, each time one or more of the processing parameters is changed during the preprocessing process, the difference between the predicted dynamic preprocessing process response and the measured dynamic preprocessing process response is used to A typical estimation error can be determined.

  In addition, the dynamic estimation error in the pre-processing is defined as an operational threshold established by one or more rules in the BIST table each time a new pre-processing dynamic estimation error is determined during the pre-processing process. Can be compared. When the dynamic estimation error in preprocessing is within the operating threshold established by one or more rules in the BIST table, the preprocessing process can continue. In this case, the heat treatment system is operating within the operating limits of the pretreatment process.

  The dynamic estimation error in the preprocessing determined for the preprocessing process is a warning threshold when the dynamic estimation error in the preprocessing is not within the operating threshold established by one or more rules in the BIST table. Can be compared. For example, a warning message may be sent when the error exceeds the operating limit but not the warning limit, and a fault message may be sent when the error exceeds the operating limit and also exceeds the warning limit. .

  In one embodiment, a warning message is sent when the dynamic estimation error in preprocessing is within a warning threshold established by one or more rules in the BIST table, but the preprocessing process continues. Is possible. In another embodiment, a warning message is sent and the preprocessing process pauses when the dynamic estimation error in the preprocessing is within a warning threshold established by one or more rules in the BIST table. Be made. In this case, the heat treatment system is operating outside the operating limits in the pretreatment process but within the warning limits. This can occur, for example, when the part is degrading or when the process is drifting. The preprocessing process can be resumed when a continue processing message is received.

  In one embodiment, when the dynamic estimation error in the preprocessing determined for the preprocessing is not within the warning threshold established by one or more rules in the BIST table, a failure message is sent and the preprocessing The process is stopped. In this case, the heat treatment system is operating outside the operating limits and outside the warning limits in the pretreatment process. This can occur, for example, when a component fails or when the system is not set up properly. The pre-processing process can be restarted when a restart message is received.

  During the pretreatment process, the pressure in the treatment chamber can be changed from a first pressure to an operating pressure. For example, if the chamber is not properly sealed, a chamber pressure error can occur. Also, during the pretreatment process, the treatment chamber temperature can be changed from a first temperature to an operating temperature. For example, if the heater is faulty, a chamber temperature error can occur. Furthermore, the chemistry within the chamber can be changed during pretreatment. For example, if a gas supply system component fails, chamber chemistry errors can occur.

  At step 1020, one or more thermal wafer processing processes are performed on a group of wafers disposed in the processing chamber. The wafer groups are arranged at different heights in the processing chamber, and a heat treatment process is performed. The wafer group may include a product wafer, a wafer with measurement means, a test wafer or a dummy wafer, or a combination thereof. For example, the thermal process can be a film formation process and / or a deposition process, and a process gas is introduced into the processing chamber. In other examples, different processes may be performed.

  The thermal processing system obtains operational data and uses the operational data to construct one or more recipes to use when processing the wafer. The operational data may also include dynamic modeling information and / or static modeling information for predicting the performance of the thermal processing system during the process. These data may further include data measured and / or predicted from previous runs.

  During wafer processing, a process may be performed in which one or more of the processing parameters are changed from the first operating value to the second operating value or maintained at the operating value.

  During wafer processing, a real-time dynamic model is executed to generate a predicted dynamic wafer process response. For example, a real-time dynamic model for wafer processing is a predictive behavior when one or more of the processing parameters are changed from a first value to a second value or maintained at an operating value during the process. Can be implemented to produce a typical wafer process response. This dynamic wafer model can be executed before, during or after the process is executed. When the process is run multiple times, one or more different wafer process dynamic models may be run.

  A dynamic wafer process response with one or more measurements is created during wafer processing. A dynamic wafer process response by measurement can be created each time one or more of the processing parameters is changed from a first operating value to a second operating value during the process. When the process is run multiple times, a dynamic wafer process response with one or more different measurements can be created.

  Also, every time one or more of the processing parameters are changed or maintained during the process, the difference between the predicted dynamic wafer process response and the measured dynamic wafer process response is used to An estimated error can be determined.

  Furthermore, the dynamic estimation error in wafer processing is compared with the operating threshold established by one or more rules in the BIST table each time a dynamic estimation error in a new wafer processing is determined during wafer processing. Can be done. Wafer processing can continue when the dynamic estimation error in wafer processing is within the operating threshold established by one or more rules in the BIST table. In this case, the heat treatment system is operating within the operating limits in wafer processing.

  A dynamic estimation error in wafer processing can be compared to a warning threshold when the dynamic estimation error in wafer processing is not within the operating threshold established by one or more rules in the BIST table. . For example, a warning message may be sent when the error exceeds the operating limit but not the warning limit, and a fault message may be sent when the error exceeds the operating limit and also exceeds the warning limit. .

  In one embodiment, when the dynamic estimation error in wafer processing is within a warning threshold established by one or more rules in the BIST table, a warning message is sent but wafer processing continues. It is possible. In another embodiment, when a dynamic estimation error in wafer processing is within a warning threshold established by one or more rules in the BIST table, a warning message is sent and the wafer processing is paused. It is done. In this case, the heat treatment system is operating outside the operational limits in wafer processing but within the warning limits. This can occur, for example, when the part is degrading, when the chamber needs cleaning, or when the process is drifting. Wafer processing can be resumed when an instruction is received.

  In one embodiment, when the dynamic estimation error in wafer processing is not within the warning threshold established by one or more rules in the BIST table, a fault message is sent and wafer processing is stopped. In this case, the heat treatment system is operating outside the operational limit and the warning limit in wafer processing. This can occur, for example, when a passive component, active component or software element fails. Wafer processing can be restarted when a restart message is received.

  During wafer processing, the pressure in the processing chamber can be maintained within operating limits established by the process recipe. For example, if a leak occurs in the chamber, or if an exhaust system component fails, a chamber pressure error may occur. Also, during wafer processing, the processing chamber temperature can be maintained within operating limits established by the process recipe. For example, a chamber temperature error can occur if a power source, heater element, or sensor fails. Further, during wafer processing, the chemistry within the chamber can be maintained within operating limits established by the process recipe. For example, if a gas supply system component fails, chamber chemistry errors can occur.

  At step 1030, one or more post-processing processes are performed while the group of wafers is still placed in the processing chamber. Post-processing processes are used, for example, to prepare for removing a group of wafers from the processing chamber. In other examples, the post-processing process may include removing a group of wafers from the processing chamber.

  The thermal processing system obtains post-processing data and uses the post-processing data to construct one or more recipes for use in post-processing the wafer. The post-processing data may also include dynamic modeling information and / or static modeling information for predicting the performance of the thermal processing system during the post-processing process. These data may further include data measured and / or predicted from previous runs.

  During the post-processing process, a process may be performed in which one or more of the processing parameters are changed from a first value, for example an operating value, to a second value.

  During the post-processing process, a real-time dynamic model is executed to generate a predicted dynamic post-processing process response. A real-time dynamic model for post-processing generates a predictive dynamic post-processing process response when one or more of the processing parameters are changed from a first value to a second value during the post-processing process Can be executed to. This post-processing dynamic model can be executed before, during or after the post-processing process is executed. When the post-processing process is executed multiple times, one or more different post-processing dynamic models can be executed.

  A dynamic post-processing process response with one or more measurements is created during the post-processing process. A dynamic post-processing process response by measurement can be created each time one or more of the processing parameters are changed from a first value to a second value during the post-processing process. When the post-processing process is performed multiple times, a dynamic post-processing process response with one or more different measurements can be created.

  Also, every time one or more of the processing parameters is changed during the post-processing process, the difference between the dynamic post-processing response by prediction and the dynamic post-processing response by measurement is used to determine the behavior in post-processing. A typical estimation error can be determined. When the dynamic estimation error in post-processing is within the operating threshold established by one or more rules in the BIST table, the post-processing process can continue. In this case, the heat treatment system is operating within operating limits in the post-treatment process.

  The dynamic estimation error in post processing can be compared to the warning threshold when the dynamic estimation error in post processing is not within the operating threshold established by one or more rules in the BIST table. . For example, a warning message may be sent when the error exceeds the operating limit but not the warning limit, and a fault message may be sent when the error exceeds the operating limit and also exceeds the warning limit. .

  In one embodiment, a warning message is sent when the dynamic estimation error in post-processing is within a warning threshold established by one or more rules in the BIST table, but the post-processing process continues. Is possible. In another embodiment, a warning message is sent and the post-processing process pauses when the dynamic estimation error in post-processing is within a warning threshold established by one or more rules in the BIST table. Be made. In this case, the heat treatment system is operating outside the operating limits in the post-processing process but within the warning limits. This can occur, for example, when the part is degrading or when the process is drifting. The post-processing process can be resumed when an instruction is received.

  In one embodiment, when the dynamic estimation error in post-processing is not within the warning threshold established by one or more rules in the BIST table, a fault message is sent and the post-processing process is stopped. In this case, the heat treatment system is operating outside the operational limits and outside the warning limits in the post-processing process. This can occur, for example, when a component fails. The pre-processing process can be restarted when a restart command is received.

  During the post-treatment process, the pressure in the treatment chamber can be changed from the operating pressure to another pressure. For example, a chamber pressure error can occur if the chamber is not properly sealed or is opened too quickly. Also, during the post-processing process, the processing chamber temperature can be changed from the operating temperature to another temperature. For example, a chamber temperature error can occur if the heating element and / or the cooling element is faulty. Furthermore, the chemistry within the chamber can be changed during post-processing. For example, if a gas supply system component fails, chamber chemistry errors can occur.

  Further, after or during the post-processing process, the plurality of wafers described above are removed from the processing chamber of the thermal processing system.

  When the process parameter is changed from the first value to the second value, the change occurs in a series of steps. One or more real-time dynamic models may be executed to generate a dynamic process response with step-by-step prediction. One or more measured dynamic process responses are created for each stage, and the difference between the predicted dynamic process response and the measured dynamic process response is used to determine a set of dynamic estimation errors Can be done. These dynamic estimation errors can be compared to an operational threshold established by one of the rules in the BIST table. When one or more of these dynamic estimation errors are not within the operating threshold established by at least one of the rules in the BIST table, the process is stopped and one or more of these dynamic estimation errors are The process may continue when it is within an operating threshold established by at least one of the rules in the BIST table.

  In one embodiment, when the process is paused because the dynamic estimation error is not within the operating threshold established by at least one of the rules in the BIST table, the dynamic process response by prediction, by measurement The real-time dynamic model used to generate the dynamic process response and / or the dynamic estimation error can be examined and a new BIST rule with an operational threshold based on the dynamic estimation error can be constructed . This new BIST rule is also entered into the BIST table and the process continues. Also, when a new BIST rule with an action threshold based on a dynamic estimation error cannot be created, the process is stopped. In other examples, the process may continue.

  In another embodiment, when the process is paused because the dynamic estimation error is not within the operating threshold established by at least one of the rules in the BIST table, a new real-time dynamic model is created. Selected and used to generate a dynamic process response with a new prediction. The new dynamic estimation error is compared to an operational threshold established by one or more rules in the BIST table. When the new dynamic estimation error is not within the operating threshold established by at least one of the rules in the BIST table, the process is stopped and the new dynamic estimation error is determined by the rule in the BIST table. The process continues when it is within the operating threshold established by at least one of the following.

  When the process is paused because the new dynamic estimation error is not within the operating threshold established by at least one of the rules in the BIST table, the dynamic process response by the new prediction, the motion by the measurement A new real-time dynamic model used to generate a dynamic process response and / or a new dynamic estimation error is examined and a new BIST with an operating threshold based on the new dynamic estimation error Rules can be constructed. This new BIST rule is also entered into the BIST table and the process continues. Also, the process is stopped when a new BIST rule with an action threshold based on a new dynamic estimation error cannot be created. In other examples, the process may continue.

  The predicted dynamic process response set can be a predicted dynamic thermal profile, predicted dynamic chamber pressure, predicted dynamic gas flow, predicted chamber chemistry, or predicted processing time, or You can have a combination of these.

  When the dynamic process response by measurement includes pressure, a pressure profile can be constructed with respect to the chamber volume. In other examples, the pressure profile may be constructed with respect to the processing volume. The volume can be divided into multiple temperature control compartments and a substantially equal pressure can be established for all temperature control compartments.

としてモデル化され得る。ただし、ｇ_１乃至ｇ_ｎは各々、チャンバー圧以外のプロセスパラメータであり、νは百分率で測定されるバルブ角度開口率であり、そしてｐはｍＴｏｒｒ単位で測定される処理チャンバー圧である。 In one example, the chamber pressure rate of change “p dots” is:

Can be modeled as However, g ₁ to g _n each is the process parameters outside the chamber pressure or, [nu is the valve angle opening ratio is measured as a percentage, and p is the processing chamber pressure is measured in mTorr units.

  When the dynamic process response by measurement includes the chamber temperature, a temperature profile can be constructed with respect to the chamber volume. In other examples, the temperature profile may be constructed with respect to the processing volume. The volume can be divided into multiple temperature control compartments, and different temperatures can be established for one or more of these temperature control compartments. In other examples, substantially equal temperatures may be established for all temperature control sections. For example, a temperature profile can be constructed based on historical data for that type of wafer, and the historical data can include a reference temperature profile.

としてモデル化され得る。ただし、ｇ_１乃至ｇ_ｎは各々、チャンバー温度以外のプロセスパラメータであり、ｈは百分率でのヒータ電力を表し、そしてＴは摂氏で測定される処理チャンバー温度である。 In another example, the rate of change of chamber temperature “T dot” is

Can be modeled as However, g ₁ to g _n each is a process parameter other than the chamber temperature, h denotes the heater power as a percentage, and T is the process chamber temperature measured in degrees Celsius.

  When the dynamic process response by measurement includes the flow rate entering and / or passing through the chamber, a flow profile can be constructed with respect to the chamber volume. In other examples, the flow profile may be constructed with respect to the processing volume. The volume can be divided into multiple compartments, and different flow models can be built for one or more of these compartments. For example, a flow profile can be constructed based on historical data for that type of process, and the historical data can include a reference flow profile.

としてモデル化され得る。ただし、ｇ_１乃至ｇ_ｎは各々、流量以外のプロセスパラメータであり、ｐはｍＴｏｒｒ単位で測定される処理チャンバー圧であり、νは百分率で測定されるバルブ角度開口率であり、そしてＲは反応物の濃度を表す。 In another example, the flow rate change rate “R dot” is:

Can be modeled as However, g ₁ to g _n are each a process parameter other than flow rate, p is the processing chamber pressure is measured in mTorr units, [nu is the valve angle opening ratio is measured as a percentage, and R is the reaction Represents the concentration of the object.

  When the dynamic process response by measurement includes wafer temperature, a temperature profile can be built for each wafer in the chamber. In other examples, the temperature profile may not be built for each wafer in the chamber. The volume can be divided into multiple temperature control zones and different temperatures can be established for wafers in different temperature control zones. In other examples, substantially equal temperatures may be assumed for all wafers. For example, a temperature profile can be constructed based on historical data for that type of wafer, and the historical data can include a reference temperature profile.

  FIG. 11 is a simplified flow diagram illustrating a method for creating a built-in self test (BIST) table for real-time monitoring of a thermal processing system, in accordance with an embodiment of the present invention.

  A BIST table can be constructed for heat treatment systems, temperature control components, pressure control components, gas supply components, machine components, computational components or software elements, or combinations thereof.

  When creating a BIST table, a process may be performed in which one or more of the processing parameters are changed from a first value to a second value. For example, the first value and / or the second value can be selected to build a normal condition, a warning condition, or a fault condition that occurs during the pre-processing process. In other examples, the first value and / or the second value may be selected to exaggerate and / or amplify errors during the preprocessing process.

BIST rules are
“Error ≦ Operational threshold (OT)” defines the operating state;
“OT <error ≦ warning threshold (WT)” defines the warning condition;
“WT <error” defines the failure condition;
Can be structured as follows. In other examples, other structures may be used.

  In order to establish operational and warning thresholds that are expected to occur during one or more processes, a number of processes are run multiple times and a set of expected dynamic error conditions and their associated operational and warning thresholds. Are used to create a set of BIST rules.

  A set of processes can be used to characterize system performance when the thermal processing system is operating within operating limits during the process. In this case, the dynamic estimation error is smaller than the operation threshold. For example, many processes are performed many times to establish an operating threshold. These processes can be performed using process parameters within operating limits. These processes may produce a set of expected errors during “normal” operation and a set of expected errors that can be used to establish operating thresholds. For example, the operational threshold may be large enough to allow some process change. In order to minimize the number of processes required to generate the operating threshold, it is possible to use the DOE method (experimental design method).

  Another set of processes can be used to characterize system performance when the heat treatment system is operating slightly outside the operating limits during the process. In these cases, the dynamic estimation error is greater than the operational threshold and smaller than the warning threshold. For example, the second set of processes is performed multiple times to establish a warning threshold. These processes can be performed using one or more process parameters slightly outside the operating limits. These processes generate an expected set of errors when the system performance deviates slightly from “normal” operation, and this set of errors is used to establish a warning threshold. For example, the warning threshold may be large enough to allow some process variation (drift) and / or component degradation, and small enough to ensure that high quality wafers are produced. Also, the warning threshold can be used to predict and prevent the occurrence of a failure. In order to minimize the number of processes required to generate the warning threshold, the DOE method (design of experiment) can be used.

  At step 1110, one or more pre-processing processes are performed on a group of wafers disposed within a processing chamber of the thermal processing system. The wafer groups are arranged at different heights in the processing chamber, and the processing chamber is sealed. The wafer group may include a product wafer, a wafer with measurement means, a test wafer or a dummy wafer, or a combination thereof. In other examples, there need not be multiple wafers.

  The heat treatment system obtains pretreatment data and uses the pretreatment data to construct one or more recipes for use in the pretreatment process. The pretreatment data may also include dynamic modeling information and / or static modeling information for predicting the performance of the heat treatment system during the pretreatment process. These data may further include data measured and / or predicted from previous runs. In the preprocessing process, an operating state is established. For example, chamber pressure, chamber temperature, substrate temperature, and / or process gas conditions are changed to operating values.

  When creating a BIST rule for a pre-processing process, a dynamic pre-processing process response by measurement can be created while pre-processing multiple wafers. In one embodiment, one or more process parameters are dynamically changed from a first value to a second value at a first rate of change. These values remain within the operating limits established for the pretreatment process. In other examples, different rates of change may be used.

  Real time for preprocessing so as to generate a predicted dynamic preprocessing process response for a plurality of wafers when one or more process parameters are dynamically changed from a first value to a second value. A dynamic model is executed.

  With respect to the preprocessing process, the dynamic estimation error in the preprocessing can be determined using the difference between the dynamic preprocessing process response by prediction and the dynamic preprocessing process response by measurement. Furthermore, the dynamic estimation error in the preprocessing is examined to determine if it can be associated with a BIST rule in the BIST table.

  When a dynamic estimation error in preprocessing cannot be associated with a BIST rule in the BIST table, a new BIST rule is created for the preprocessing process, and a new BIST rule cannot be created The pretreatment process can be stopped.

  In addition, a new operating threshold is established for the new BIST rule. In one embodiment, the new behavior threshold for the new BIST rule is based on the behavior threshold used to compare the dynamic estimation error in the preprocessing determined for the preprocessing. In other examples, the new operating threshold may be determined independently of it.

  In one embodiment, a new warning threshold is established based on the new operational threshold. In other examples, the new warning threshold may be determined independently of it.

  A warning message is created and associated with the new BIST rule. For example, this warning message may be sent when the dynamic estimation error in the pre-processing is outside the operating threshold but is within the new warning threshold established for the new BIST rule. A failure message is also created and associated with the new BIST rule. For example, this failure message may be sent when the dynamic estimation error in preprocessing is not within the new warning threshold established for the new BIST rule.

  In addition, during the preprocessing process used to establish the dynamic estimation error in the new preprocessing and associated with the new BIST rule, a new predictive dynamic preprocessing process response for multiple wafers. To generate, a new real-time dynamic model for preprocessing can be executed. Then, during the pre-processing process used to establish the dynamic estimation error in the new pre-processing and associated with the new BIST rule, the dynamic pre-processing process response due to the new measurement for multiple wafers Produced.

  During the pretreatment process, the pressure in the treatment chamber can be dynamically changed from a first pressure to an operating pressure. For example, BIST rules can monitor chamber pressure errors and a warning or fault message can be sent when the chamber is not properly sealed. Also, during the pretreatment process, the treatment chamber temperature can be dynamically changed from the first temperature to the operating temperature. For example, BIST rules can monitor chamber temperature errors and a warning or fault message can be sent when the heater is faulty. Furthermore, the chemistry within the chamber can be altered during the pretreatment process. For example, BIST rules can monitor chamber chemistry errors and a warning or fault message can be sent when a gas supply system component has failed.

  Rule creation can be terminated when all of the dynamic estimation errors in the preprocessing determined during the preprocessing process are within the operational threshold established by one or more rules in the BIST table. When the dynamic estimation error in a pre-process is not within the operational threshold established by one or more rules in the BIST table, rule creation continues and one or more new rules can be created.

  At step 1120, one or more wafer processing processes are performed while the group of wafers is placed in the processing chamber. The wafer groups are arranged at different height positions in the processing chamber, and a thermal process is performed. The wafer group may include a product wafer, a wafer with measurement means, a test wafer or a dummy wafer, or a combination thereof. For example, the thermal process can be a film formation process and / or a deposition process. In other examples, different processes may be performed.

  The thermal processing system obtains wafer processing data and uses the data to build one or more recipes to use when processing the wafer. These data may also include dynamic modeling information and / or static modeling information for predicting the performance of the thermal processing system during the thermal process. These data may further include data measured and / or predicted from previous runs.

  During wafer processing, a process may be performed in which one or more of the processing parameters are changed from a first value to a second value or maintained at an operating value. When building a BIST table, the first value and / or the second value, or operational value, can be selected to generate a warning condition and / or a failure condition during wafer processing. In other examples, the first value and / or the second value may be selected to exaggerate and / or amplify errors during wafer processing.

  When creating BIST rules for a wafer processing process, a dynamic wafer process response by measurement can be created while wafer processing multiple wafers. In one embodiment, at least one process parameter is maintained at an operating value during the wafer processing process. In another example, one or more process parameters are dynamically changed from a first value to a second value at a first rate of change. These values remain within the operating limits established by the process recipe of the wafer processing process.

  When one or more process parameters are dynamically changed from a first value to a second value or maintained at an operating value, to generate a predicted dynamic wafer process response for a plurality of wafers A real-time dynamic model for wafer processing is executed.

  For wafer processing processes, a dynamic estimation error in wafer processing can be determined using the difference between the predicted dynamic wafer process response and the measured dynamic wafer process response. In addition, dynamic estimation errors in wafer processing are examined to determine if it can be associated with BIST rules in the BIST table.

  When a dynamic estimation error in wafer processing cannot be associated with a BIST rule in the BIST table, a new BIST rule is created for the wafer processing process, and a new BIST rule cannot be created The wafer processing process can be stopped.

  In addition, a new operating threshold is established for the new BIST rule. In one embodiment, the new operational threshold for the new BIST rule is based on the operational threshold used to compare dynamic estimation errors in wafer processing determined for the wafer processing process. In other examples, the new operating threshold may be determined independently of it.

  In one embodiment, a new warning threshold is established based on the new operational threshold. In other examples, the new warning threshold may be determined independently of it.

  A warning message is created and associated with the new BIST rule. For example, this warning message may be sent when the dynamic estimation error in wafer processing is outside the operating threshold but is within the new warning threshold established for the new BIST rule. A failure message is also created and associated with the new BIST rule. For example, this fault message may be sent when the dynamic estimation error in wafer processing is not within the new warning threshold established for the new BIST rule.

  In addition, a new predictive dynamic wafer process response is generated for multiple wafers during the wafer processing process used to establish dynamic estimation errors in new wafer processing and associated with new BIST rules. To do so, a real-time dynamic model for new wafer processing can be implemented. And during the wafer processing process used to establish dynamic estimation errors in new wafer processing and associated with new BIST rules, a dynamic wafer process response with new measurements is created for multiple wafers. It is.

  During the wafer processing process, the pressure in the processing chamber can be dynamically changed from a first pressure to an operating pressure. For example, BIST rules can monitor chamber pressure errors and a warning or fault message can be sent when the chamber is not properly sealed. Also, during the wafer processing process, the processing chamber temperature can be dynamically changed from the first temperature to the operating temperature. For example, BIST rules can monitor chamber temperature errors and a warning or fault message can be sent when the heater is faulty. Furthermore, the chemistry within the chamber can be changed during the wafer processing process. For example, BIST rules can monitor chamber chemistry errors and a warning or fault message can be sent when a gas supply system component has failed.

  In order to establish an operating threshold for a set of error conditions that are expected to occur during one or more wafer processing processes, a number of wafer processing processes are performed many times, and this expected set of error conditions and their A set of BIST rules is created using the thresholds for. Furthermore, in order to establish a set of operational thresholds for error conditions that are unlikely to occur during the wafer processing process, the wafer processing process is performed a number of times, using this set of unlikely error conditions and their associated thresholds. A set of BIST rules is created.

  Rule creation can be terminated when all of the dynamic estimation errors in wafer processing determined during the wafer processing process are within the operating threshold established by one or more rules in the BIST table. When the dynamic estimation error in a wafer process is not within the operating threshold established by one or more rules in the BIST table, rule creation continues and one or more new rules can be created.

  At step 1130, one or more post-processing processes are performed while the wafer group is still placed in the processing chamber. Post-processing processes are used, for example, to prepare for removing a group of wafers from the processing chamber. In other examples, the post-processing process may include removing a group of wafers from the processing chamber.

  The thermal processing system obtains post-processing data and uses the post-processing data to construct one or more recipes for use in post-processing the wafer. The post-processing data may also include dynamic modeling information and / or static modeling information for predicting the performance of the thermal processing system during the post-processing process. These data may further include data measured and / or predicted from previous runs.

  When creating BIST rules for a post-processing process, a dynamic post-processing process response by measurement can be created while post-processing multiple wafers. In one embodiment, one or more process parameters are dynamically changed from a first value to a second value at a first rate of change. These values remain within the operating limits established for the post-processing process. In other examples, different rates of change may be used.

  Real time for post processing to generate a predicted dynamic post processing process response for a plurality of wafers when one or more process parameters are dynamically changed from a first value to a second value. A dynamic model is executed.

  With respect to the post-processing process, the dynamic estimation error in the post-processing can be determined using the difference between the dynamic post-processing process response by prediction and the dynamic post-processing process response by measurement. Further, the dynamic estimation error in post processing is examined to determine if it can be associated with a BIST rule in the BIST table.

  When a dynamic estimation error in post-processing cannot be associated with a BIST rule in the BIST table, a new BIST rule is created for the post-processing process, and a new BIST rule cannot be created The post-processing process can be stopped.

  In addition, a new operating threshold is established for the new BIST rule. In one embodiment, the new behavior threshold for the new BIST rule is based on the behavior threshold used to compare the dynamic estimation error in the post processing determined for post processing. In other examples, the new operating threshold may be determined independently of it.

  In one embodiment, a new warning threshold is established based on the new operational threshold. In other examples, the new warning threshold may be determined independently of it.

  A warning message is created and associated with the new BIST rule. For example, this warning message may be sent when the dynamic estimation error in the post-processing is outside the operating threshold but within the new warning threshold established for the new BIST rule. A failure message is also created and associated with the new BIST rule. For example, this failure message may be sent when the dynamic estimation error in post-processing is not within the new warning threshold established for the new BIST rule.

  In addition, during the post-processing process used to establish the dynamic estimation error in the new post-processing and associated with the new BIST rule, the dynamic post-processing response with the new prediction for multiple wafers. To generate, a new post-processing real-time dynamic model can be executed. Then, during the post-processing process used to establish the dynamic estimation error in the new post-processing and associated with the new BIST rule, the dynamic post-processing process response due to the new measurement for multiple wafers Produced.

  During the post-treatment process, the pressure in the treatment chamber can be dynamically changed from operating pressure to another pressure. For example, BIST rules can monitor chamber pressure errors and a warning or fault message can be sent when the chamber is not properly sealed. Also, during the post-processing process, the processing chamber temperature can be dynamically changed from operating temperature to another temperature. For example, BIST rules can monitor chamber temperature errors and a warning or fault message can be sent when the heater is faulty. Furthermore, the chemistry within the chamber can be changed during the post-treatment process. For example, BIST rules can monitor chamber chemistry errors and a warning or fault message can be sent when a gas supply system component has failed.

  Rule creation can be terminated when all of the dynamic estimation errors in the post-processing are within an operational threshold established by at least one rule in the BIST table. When the dynamic estimation error in a post-processing is not within the operating threshold established by at least one rule in the BIST table, rule creation continues and one or more new rules can be created.

  Real and / or virtual sensors can be used to create a dynamic process response by measurement. For example, the dynamic process response by measurement of the processing chamber can include measuring the temperature of each of the plurality of temperature control compartments. In other examples, per-part measurements may not be required. In other embodiments, optical techniques may be used to measure the temperature in the chamber and / or the wafer temperature.

  The present invention has been illustrated by the description of various embodiments, and these embodiments have been described in considerable detail. However, it is not intended that the appended claims be limited in any way to such detailed forms. Those skilled in the art will readily realize further advantages and modifications. The invention in its broader aspects is therefore not limited to the specific details, exemplary apparatus and methods, and illustrated embodiments shown and described herein. Accordingly, these details may be changed without departing from the scope of the general inventive concept of the applicant.

1 is a perspective view schematically showing a heat treatment system according to an embodiment of the present invention. It is a figure which cuts and shows a part of heat treatment system according to an embodiment of the present invention. 1 is a simplified block diagram illustrating a heat treatment system according to an embodiment of the present invention. FIG. 6 is a diagram illustrating a transient behavior of chamber pressure with respect to a first pressure set point change, according to an embodiment of the present invention. FIG. 6 is a diagram illustrating the transient behavior of chamber pressure with respect to a second pressure set point change, in accordance with an embodiment of the present invention. FIG. 6 is a diagram illustrating the transient behavior of chamber pressure with respect to a third pressure set point change, in accordance with an embodiment of the present invention. FIG. 6 shows the rate of increase as a function of gas flow rate for a three stage response test case according to an embodiment of the present invention. FIG. 4 shows chamber pressure at various valve angles as a function of flow rate according to an embodiment of the invention. FIG. 6 schematically illustrates one embodiment of a dynamic model characterizing one or more of the responses of a heat treatment system according to an embodiment of the present invention. FIG. 6 is a simplified flow diagram illustrating a method for monitoring a heat treatment system in real time using a built-in self test (BIST) table, in accordance with an embodiment of the present invention. FIG. 3 is a simplified flow diagram illustrating a method for creating a built-in self test (BIST) table for real-time monitoring of a heat treatment system, in accordance with an embodiment of the present invention.

Claims (50)

A method for monitoring a heat treatment system in real time using a built-in self test (BIST) table:
Disposing a plurality of wafers at different heights in a processing chamber of the heat treatment system;
Creating a dynamic pretreatment process response by measurement while subjecting the plurality of wafers to a pretreatment process, wherein at least one first process parameter is an operation established for the pretreatment process. Being changed from a first value to a second value within limits;
Preprocessing to generate a predictive dynamic preprocessing process response for the plurality of wafers when the at least one first process parameter is changed from the first value to the second value. Running a real-time dynamic model for
Determining a dynamic estimation error in preprocessing for the preprocessing process using a difference between the dynamic preprocessing process response from the prediction and the dynamic preprocessing process response from the measurement;
Comparing the dynamic estimation error in the pre-processing with one or more first action thresholds established by one or more first rules in the BIST table;
Continuing the preprocessing process when a dynamic estimation error in the preprocessing is within the one or more first motion thresholds; and a dynamic estimation error in the preprocessing is the one or more Suspending the pretreatment process when not within the first operating threshold of
Having a method. Creating a dynamic wafer process response by measurement while subjecting the plurality of wafers to a wafer processing process, wherein at least one second process parameter is maintained at an operating value in the wafer processing process;
Performing a real-time dynamic model for wafer processing to generate a predicted dynamic wafer process response for the plurality of wafers during the wafer processing process;
Determining a dynamic estimated error in wafer processing for the wafer processing process using a difference between the predicted dynamic wafer process response and the measured dynamic wafer process response;
Comparing the dynamic estimation error in the wafer processing with one or more second operating thresholds established by one or more second rules in the BIST table;
Allowing the wafer processing process to continue when a dynamic estimation error in the wafer processing is within the one or more second operating thresholds; and a dynamic estimation error in the wafer processing is the one or more Suspending the wafer processing process when not within the second operating threshold of
The method of claim 1 further comprising: Creating a dynamic post-processing response by measurement while subjecting the plurality of wafers to a post-processing process, wherein at least one third process parameter is varied from an operating value in the post-processing process;
Executing a real-time dynamic model for post-processing during the post-processing process to generate a predicted dynamic post-processing process response for the plurality of wafers;
Determining a dynamic estimation error in post-processing for the post-processing process using a difference between the dynamic post-processing process response by the prediction and the dynamic post-processing process response by the measurement;
Comparing the dynamic estimation error in the post-processing with one or more third action thresholds established by one or more third rules in the BIST table;
Allowing the post-processing process to continue when the dynamic estimation error in the post-processing is within the one or more third motion thresholds; and the dynamic estimation error in the post-processing is the one or more Suspending the post-processing process when not within the third operating threshold of
The method of claim 2 further comprising: When the dynamic estimation error in the preprocessing is not within the range of the one or more first motion thresholds:
Comparing the dynamic estimation error in the preprocessing with a warning threshold established by the one or more first rules in the BIST table;
Sending a warning message and allowing the pre-processing process to continue when a dynamic estimation error in the pre-processing is within the warning threshold; and a dynamic estimation error in the pre-processing is the warning threshold Sending a failure message and stopping the pre-processing process when not within the range;
The method of claim 1 further comprising: A dynamic estimation error in the wafer processing when a dynamic estimation error in the wafer processing is not within the one or more second operating thresholds established by the one or more second rules in the BIST table. Comparing the estimated error to a warning threshold established by the one or more second rules in the BIST table;
Sending a warning message and allowing the wafer processing process to continue when a dynamic estimation error in the wafer processing is within the warning threshold; and a dynamic estimation error in the wafer processing is the warning threshold When not in range, send a fault message and stop the wafer processing process;
The method of claim 2 further comprising: When the dynamic estimation error in the post-processing is not within the range of the one or more third action thresholds established by the one or more third rules in the BIST table, the dynamic in the post-processing Comparing the estimated error to a warning threshold established by the one or more third rules in the BIST table;
Sending a warning message and continuing the post-processing process when a dynamic estimation error in the post-processing is within the warning threshold; and a dynamic estimation error in the post-processing is the warning threshold When not in range, send a fault message and stop the post-processing process;
The method of claim 3 further comprising: When the dynamic estimation error in the preprocessing is not within the range of the one or more first action thresholds established by the one or more first rules in the BIST table;
Creating a new first BIST rule having a new first operation threshold based on a dynamic estimation error in the preprocessing determined for the plurality of wafers during the preprocessing process; Allowing the preprocessing process to continue after is entered into the BIST table;
The method of claim 1 further comprising: Establishing at least one warning threshold for the new first BIST rule;
Creating a warning message to be associated with the new first BIST rule; and creating a failure message to be associated with the new first BIST rule;
The method of claim 7 further comprising: When a dynamic estimation error in the wafer processing is not within the range of the one or more second operating thresholds established by the one or more second rules in the BIST table;
Creating a new second BIST rule having a new second operation threshold based on a dynamic estimation error in the wafer processing determined for the plurality of wafers during the wafer processing process; and the new second BIST rule Allowing the wafer processing process to continue after is entered into the BIST table;
The method of claim 2 further comprising: Establishing at least one warning threshold for the new second BIST rule;
Creating a warning message to be associated with the new second BIST rule; and creating a failure message to be associated with the new second BIST rule;
10. The method of claim 9, further comprising: When the dynamic estimation error in the post-processing is not within the range of the one or more third action thresholds established by the one or more third rules in the BIST table;
Creating a new third BIST rule having a new third operation threshold based on a dynamic estimation error in the post-processing determined for the plurality of wafers during the post-processing process; and the new third BIST rule Allowing the post-processing process to continue after is entered into the BIST table;
The method of claim 3 further comprising: Establishing at least one warning threshold for the new third BIST rule;
Creating a warning message to be associated with the new third BIST rule; and creating a failure message to be associated with the new third BIST rule;
The method of claim 11, further comprising: Each of the at least one first, second and third process parameters includes a chamber pressure, and the chamber pressure rate of change is:
g ₁ to process parameters each other than the chamber pressure g _n, [nu valve angle opening ratio measured as a percentage, p is the pressure of the processing chamber to be measured in mTorr units,

Modeled in the
The method of claim 3. Each of the at least one first, second, and third process parameters includes a chamber temperature, and the chamber temperature change rate is:
g ₁ to g _n each process parameters other than the chamber temperature, [nu valve angle opening ratio measured in percent, as the temperature of the processing chamber T is measured in ℃ units,

Modeled in the
The method of claim 3. Selecting a new, pre-processing real-time dynamic model for use during the pre-processing process;
Executing the new pre-processing real-time dynamic model to generate a new predictive dynamic pre-processing process response for the plurality of wafers during the pre-processing process;
Determining a new, dynamic estimation error in the pre-processing using a difference between the dynamic pre-processing process response by the new prediction and the dynamic pre-processing process response by the measurement;
Comparing the dynamic estimation error in the new pre-processing with the one or more first action thresholds established by the one or more first rules in the BIST table;
Continuing the preprocessing process when a dynamic estimation error in the new preprocessing is within the one or more first motion thresholds; and a dynamic estimation error in the new preprocessing is Pausing the pretreatment process when not within the one or more first operating thresholds;
The method of claim 1 further comprising: Create a new first BIST rule having a new first action threshold based on a dynamic estimation error in the new preprocessing, and continue the preprocessing process after the new first BIST rule is created Or stopping the preprocessing process when the new first BIST rule cannot be created;
16. The method of claim 15, further comprising: Selecting a new real-time dynamic model for wafer processing to be used during the wafer processing process;
Executing a real-time dynamic model for the new wafer processing to generate a new predictive dynamic wafer process response for the plurality of wafers during the wafer processing process;
Determining a new dynamic estimation error in wafer processing using the difference between the dynamic wafer process response from the new prediction and the dynamic wafer process response from the measurement;
Comparing a dynamic estimation error in the new wafer process with the one or more second operating thresholds established by the one or more second rules in the BIST table;
Allowing the wafer processing process to continue when a dynamic estimation error in the new wafer processing is within the one or more second operating thresholds; and a dynamic estimation error in the new wafer processing is Pausing the wafer processing process when not within the one or more second operating thresholds;
The method of claim 2 further comprising: A new second BIST rule having a new second operation threshold based on a dynamic estimation error in the new wafer processing is created, and the wafer processing process is continued after the new second BIST rule is created Or stopping the wafer processing process when the new second BIST rule cannot be created;
18. The method of claim 17, further comprising: Selecting a new post-processing real-time dynamic model to be used during the post-processing process;
Executing the new post-processing real-time dynamic model to generate a new predictive dynamic post-processing process response for the plurality of wafers during the post-processing process;
Determining a new post-processing dynamic estimation error using a difference between the new prediction dynamic post-processing process response and the measurement dynamic post-processing process response;
Comparing a dynamic estimation error in the new post-processing with the one or more third action thresholds established by the one or more third rules in the BIST table;
Continuing the post-processing process when a dynamic estimation error in the new post-processing is within the one or more third motion thresholds; and a dynamic estimation error in the new post-processing is Pausing the post-processing process when not within the one or more third operating thresholds;
The method of claim 3 further comprising: Create a new third BIST rule having a new third action threshold based on a dynamic estimation error in the new post-processing, and continue the post-processing process after the new third BIST rule is created Or stopping the post-processing process if the new third BIST rule cannot be created;
20. The method of claim 19, further comprising:   The method of claim 1, further comprising changing the at least one first process parameter from the first value to the second value using a series of steps.   The predicted dynamic pretreatment process response comprises a predicted dynamic temperature gradient, a predicted dynamic chamber pressure, a predicted dynamic gas flow, or a predicted processing time, or a combination thereof. Item 2. The method according to Item 1.   The processing chamber has a plurality of temperature control compartments, and creating the dynamic pretreatment process response by the measurement with respect to the processing chamber comprises measuring the temperature of each of the plurality of temperature control compartments. Item 2. The method according to Item 1. The processing chamber has a plurality of temperature control compartments, the method further comprising:
Modeling the thermal interaction between the temperature control compartments of the processing chamber; and incorporating the model of the thermal interaction into the real-time dynamic model for the pretreatment;
The method of claim 1, comprising: Modeling the thermal interaction between the processing chamber and the surrounding environment; and incorporating the model of the thermal interaction into the real-time dynamic model for the pretreatment;
The method of claim 1 further comprising: Modeling a thermal interaction between the plurality of wafers and a processing space in the processing chamber; and incorporating the model of the thermal interaction into the real-time dynamic model for the pre-processing;
The method of claim 1 further comprising:   The method of claim 1, wherein the at least one first process parameter includes a curvature of a wafer.   The method of claim 1, wherein at least one test wafer is included in the plurality of wafers. Receiving feedforward data; and using the feedforward data to determine a real-time dynamic model for the preprocessing;
The method of claim 1 further comprising: Receiving feedback data; and using the feedback data to determine a real-time dynamic model for the preprocessing;
The method of claim 1 further comprising:   The method of claim 30, wherein the feedback data comprises correction data, error data, measurement data, or historical data, or a combination of two or more thereof. Placing a plurality of wafers at different heights in a processing chamber of a heat treatment system;
Creating a dynamic pretreatment process response by measurement while subjecting the plurality of wafers to a pretreatment process, wherein at least one first process parameter is an operation established for the pretreatment process. Being changed from a first value to a second value within limits;
Preprocessing to generate a predictive dynamic preprocessing process response for the plurality of wafers when the at least one first process parameter is changed from the first value to the second value. Running a real-time dynamic model for
Determining a dynamic estimation error in preprocessing for the preprocessing process using the difference between the dynamic preprocessing process response from the prediction and the dynamic preprocessing process response from the measurement;
Comparing the dynamic estimation error in the pre-processing with one or more first operational thresholds established by one or more first rules in a built-in self test (BIST) table;
Continuing the preprocessing process when a dynamic estimation error in the preprocessing is within the one or more first motion thresholds; and a dynamic estimation error in the preprocessing is the one or more Suspending the pretreatment process when not within the first operating threshold of
A computer-readable medium storing computer-executable instructions for executing. A method of operating a controller of a thermal processing system configured to process a substrate, comprising:
Instructing the thermal processing system to place a plurality of wafers at different height positions within a processing chamber of the thermal processing system;
Instructing the thermal processing system to create a dynamic pretreatment process response by measurement while subjecting the plurality of wafers to a pretreatment process, wherein the at least one first process parameter is the Changing from a first value to a second value within operating limits established for the pretreatment process;
The thermal processing system generates a predictive dynamic pretreatment process response for the plurality of wafers when the at least one first process parameter is changed from the first value to the second value. Order to execute a real-time dynamic model for preprocessing to
Using the difference between the predicted dynamic pretreatment process response and the measured dynamic pretreatment process response to the heat treatment system, the dynamic estimation error in the pretreatment is calculated for the pretreatment process. Ordering to decide;
Instructing the heat treatment system to compare the dynamic estimation error in the pre-treatment with one or more first operating thresholds established by one or more first rules in a built-in self test (BIST) table. Stage to do;
Instructing the heat treatment system to continue the pretreatment process when a dynamic estimation error in the pretreatment is within the one or more first operating thresholds; and Instructing to pause the preprocessing process when a dynamic estimation error in the preprocessing is not within the one or more first operating thresholds;
Having a method. A method for monitoring a heat treatment system in real time using a built-in self test (BIST) table:
Disposing a plurality of wafers at different heights in a processing chamber of the heat treatment system;
Changing the process parameters from a first value to a second value within operating limits established for the process or maintaining the operating values while subjecting the plurality of wafers to the process;
Creating a dynamic process response by measurement as the process parameters are changed or maintained during the process;
Executing a real-time dynamic model to generate a predictive dynamic process response for the plurality of wafers as the process parameters are changed or maintained;
Determining a dynamic estimation error for the process using a difference between the predicted dynamic process response and the measured dynamic process response;
Comparing the dynamic estimation error to an operational threshold established by one or more rules in the BIST table;
Allowing the process to continue when the dynamic estimation error is within the operating threshold; and pausing the process when the dynamic estimation error is not within the operating threshold;
Having a method.   35. The method of claim 34, wherein the process is a pretreatment process performed prior to heat treatment of the plurality of wafers.   35. The method of claim 34, wherein the process is a thermal wafer processing process that deposits or grows a film on the plurality of wafers.   The creating step, executing step, determining step, comparing step, continuing step and suspending step are first performed by preprocessing in which the process parameter is changed from the first value to the second value. 35. The method of claim 34, wherein the method is performed in a process and then again in a thermal wafer processing process in which the process parameters are maintained at the operating values. When the dynamic estimation error is not within the operating threshold range:
Comparing the dynamic estimation error to a warning threshold established by the one or more rules in the BIST table;
Sending a warning message when the dynamic estimation error is within the warning threshold and allowing the process to continue; and when the dynamic estimation error is not within the warning threshold Sending a message and stopping the process;
35. The method of claim 34, further comprising: When the dynamic estimation error is not within the operating threshold range:
Creating a new BIST rule with a new operating threshold based on the dynamic estimation error, entering the new BIST rule into the BIST table, and then allowing the process to continue;
35. The method of claim 34, further comprising: Establishing at least one warning threshold for the new BIST rule;
Creating a warning message to be associated with the new BIST rule; and creating a failure message to be associated with the new BIST rule;
40. The method of claim 39, further comprising: The process parameter is chamber pressure and the chamber pressure change rate is:
g ₁ to process parameters each other than the chamber pressure g _n, [nu valve angle opening ratio measured as a percentage, p is the pressure of the processing chamber to be measured in mTorr units,

Modeled in the
35. The method of claim 34. The process parameter is the chamber temperature and the chamber temperature change rate is:
g ₁ to g _n each process parameters other than the chamber temperature, [nu valve angle opening ratio measured in percent, as the temperature of the processing chamber T is measured in ℃ units,

Modeled in the
35. The method of claim 34.   35. The predicted dynamic process response comprises a predicted dynamic temperature gradient, a predicted dynamic chamber pressure, a predicted dynamic gas flow, or a predicted processing time, or a combination thereof. The method described.   35. The process chamber according to claim 34, wherein the process chamber has a plurality of temperature control compartments, and creating the dynamic process response by the measurement with respect to the process chamber comprises measuring the temperature of each of the plurality of temperature control compartments. The method described. The processing chamber has a plurality of temperature control compartments, the method further comprising:
Modeling the thermal interaction between the temperature control compartments of the processing chamber; and incorporating the model of the thermal interaction into the real-time dynamic model;
35. The method of claim 34, comprising: Modeling the thermal interaction between the processing chamber and the surrounding environment; and incorporating the model of the thermal interaction into the real-time dynamic model;
35. The method of claim 34, further comprising: Modeling a thermal interaction between the plurality of wafers and a processing space in the processing chamber; and incorporating the model of the thermal interaction into the real-time dynamic model;
35. The method of claim 34, further comprising: Receiving feedforward data; and determining the real-time dynamic model using the feedforward data;
35. The method of claim 34, further comprising: Receiving feedback data; and determining the real-time dynamic model using the feedback data;
35. The method of claim 34, further comprising:   50. The method of claim 49, wherein the feedback data comprises correction data, error data, measurement data, or historical data, or a combination of two or more thereof.

Images